Who Are We Missing? Reporting of Ethnicity, Race, and Sex‐Specific Populations in Clinical Trials

Low enrollment in randomized clinical trials (RCTs) hinders cardiovascular medicine advances. ¹ , ² This disproportionately affects Black, Hispanic and Latino, and female patients who have disproportionately higher rates of cardiovascular risk factors and worse outcomes. ¹ Prior studies evaluated individual diseases, limited populations, and older trials, and used prior National Institutes of Health race and ethnicity definitions. ¹ , ² Therefore, our objective was to quantitatively evaluate Black, Hispanic and Latino, and female patients’ representativeness using US disease and risk factor prevalences across contemporary cardiovascular trials. We hypothesized that these populations would be underrepresented across trial categories.

We performed a systematic PubMed/Medline review of cardiovascular disease or risk factor RCTs published from 2017 to 2022 that enrolled >100 individual US adults following Preferred Reporting Items for Systematic reviews and Meta‐Analyses (PRISMA) US guidelines. Two reviewers (E.B.F., I.K.) screened and selected studies meeting eligibility criteria. A third reviewer (A.L.) adjudicated disagreements.

The University of Pennsylvania Institutional Review Board exempted this study (PROSPERO: CRD42023384196). We abstracted inclusion criteria, total patients, publication year, funding source, trial intent (treatment, prevention, diagnosis, safety), intervention type (medication, device, procedural, behavioral), and enrolled populations by race, ethnicity, and sex. We classified trials as enrolling patients with heart failure, coronary artery disease, type 2 diabetes, atrial fibrillation, valvular disease, stroke, hyperlipidemia, and peripheral arterial disease. To describe representativeness of Black, Hispanic and Latino, and female patients in RCTs relative to their US prevalence, we estimated participation‐to‐prevalence ratios (PPRs) averaged for each cardiovascular disease or risk factor using the American Heart Association Heart Disease and Stroke Statistics 2023 Update for prevalence data. ³ PPRs are calculated by dividing the proportion of the population of interest enrolled in each type of RCT by the corresponding proportion of US disease or risk factor prevalence. PPRs <0.8 indicate underrepresentativeness and >1.2 indicate overrepresentativeness. ² We calculated unreported race, ethnicity, and sex prevalence using reported intersectional prevalences (e.g., Black, not Hispanic or Latino, and female) and validated 2022 American Community Survey Census data. ⁴

Data collection forms, data extracted from included studies, data used for analyses, and analytic code will be made available by reasonable request to the study's principal investigator.

Among 2304 trials screened, 58 (2.5%) that cumulatively enrolled 123 734 patients met inclusion criteria. Twenty‐six trials enrolled patients with heart failure. Median enrollment across all trials was 905 participants (interquartile range [IQR], 345–2725). Seventy‐six percent were industry‐funded, 86% assessed therapeutic interventions, and 66% investigated medications. Median enrollment of Black, Hispanic and Latino, and female patients was 164 (IQR, 61–264), 40 (IQR, 16–481), and 355 (IQR, 101–893), respectively.

We estimated 22 of 24 possible PPRs ranging from 0 to 0.93: 8 among Black patients and 7 each among Hispanic and Latino and female patients. Black patients in hyperlipidemia trials, Hispanic and Latino patients in heart failure trials, and female patients in stroke trials were representative. No studies had PPRs >1, nor >1.2 indicating overrepresentativeness (Figure).

Figure . Participation‐to‐prevalence ratios.

The marker sizes correspond to the proportion of Black, Hispanic and Latino, and female patients for each cardiovascular disease or risk factor trial category comprised of the overall respective populations (Black patients: 7525 overall; Hispanic and Latino patients: 13 093 overall; female patients: 38 598 overall). Participation‐to‐prevalence ratios are calculated by dividing the proportion of the population of interest enrolled in each type of cardiovascular trial by the corresponding proportion of US disease or risk factor prevalence. Participation‐to‐prevalence ratios <0.8 indicate underrepresentativeness and >1.2 indicate overrepresentativeness. We estimated 22 of 24 possible participation‐to‐prevalence ratios (3 per trial type). The proportion of US disease prevalence was not available for Hispanic and Latino patients with atrial fibrillation and female patients with valvular disease.

Underrepresentativeness pervades cardiovascular RCTs, as we novelly quantified with 86% of PPRs (19 out of 22) demonstrating underrepresentativeness of Black, Hispanic and Latino, and female patients across diseases and risk factors. Mechanisms are likely multifactorial, including structural racism and heterogeneous and atypical disease presentation among women, leading to decreased diagnosis and health care access; increased costs of recruiting, enrolling, and retaining minoritized populations without budgetary considerations by funding agencies; not approaching, enrolling, or retaining representative populations or understanding wherein the process representativeness is most impacted; mistrust of medical systems; and a lack of RCT demographic data collection and reporting (i.e., the PPR numerator), including intersectional data. Among included studies, unknown race, not Hispanic or Latino was the second largest racial group after white, despite National Institutes of Health RCT demographic enrollment requirements.

In addition to missing PPR numerator data, cardiovascular disease and risk factor prevalences (i.e., the PPR denominator) were challenging to identify. Although the American Heart Association Statistical Update provided the most detailed data, it did not report most prevalences by race or ethnicity. Furthermore, most landmark cardiovascular trials are not exclusively performed in the US, thus limiting the generalizability to US populations. Only 15 of the 58 trials enrolled exclusively in the US, and 14 did not report the US population enrolled. However, US uptake of trial findings is widespread, making our findings particularly pertinent to Black, Hispanic and Latino, and female patients. Without more US‐specific trials and robust prevalence data, clinicians, health care administrators, and policymakers cannot understand true disease burden among underrepresented populations, hindering priority setting, developing improvement plans, and allocating funding. This is compounded by underestimates of prevalence data, because diagnosis requires medical access and does not account for clinician bias.

Importantly, 76% of studies had industry funders. To promote representative RCTs, regulators and payers should mandate and enforce adequate enrollment and retention of Black, Hispanic and Latino, and female patients within cardiovascular RCTs. For example, Food and Drug Administration industry guidance for enrolling diverse cohorts should be actively enforced. ⁵

Despite knowledge that RCT underrepresentativeness threatens generalizability, reinforces inequities in cardiovascular disease, and may contribute to medical mistrust, we demonstrated underrepresentativeness of Black, Hispanic and Latino, and female patients across all contemporary cardiovascular RCTs. To improve representativeness, future work is needed to improve RCT demographic, recruitment, payment reporting, and US prevalence data, increase US trial enrollment, and evaluate the impact of various strategies on increasing representativeness.

Sources of Funding

Support for this project is from the American Heart Association. The American Heart Association played no role in the performance or interpretation of this study.

Disclosures

Dr Lane‐Fall is a paid educational consultant for Janssen Pharmaceuticals and Medtronic.